The doctoral research program Learning for Adaptive Optimization focuses on developing learning-based methods for adaptive optimization in dynamic, uncertain, and/or complex environments. The research targets solving various complex problems where conventional optimization approaches are often ineffective or inefficient (e.g. require problem specific tuning), or where optimization objectives, constraints, or system dynamics evolve over time, rendering static methods inadequate.
The PhD candidate will explore the integration of machine learning techniques into optimization frameworks to enable continuous adaptation driven by data and feedback. Research topic may include adaptive and online optimization, reinforcement learning for decision-making, surrogate or model-based optimization, meta-learning for algorithm adaptation, and hybrid approaches that combine learning with exact or heuristic optimization methods.
The research will emphasize well-founded methodological design, analysis of convergence and robustness, and experimental validation on real-world or (benchmark) simulated problems. Application areas may include engineering systems, intelligent control, resource allocation, and complex industrial processes. The expected outcome is the development of robust, adaptive optimization methods that enhance performance, efficiency, and resilience in challenging, unpredictable, or evolving environments.